JP7318744B2 - Optimization engine, optimization method, and program - Google Patents

Description

The present disclosure relates to an optimization engine, optimization method, and program for selecting the optimum one from multiple access networks.

When users use network services, they often use communication lines provided by communication carriers. There are various physical media that can be used, such as optical lines and wireless communication lines. For example, in the case of optical lines, IEEE 802.3 (Ethernet (registered trademark)) and ITU-T G.3. 983/G. 984/G. 987/G. 989 and other communication standards exist. In the case of wireless communication lines, there are communication standards such as 3GPP 36 Series (LTE), IEEE 802.11 (wireless LAN), and IEEE 802.16 (WiMAX).

A user terminal can communicate by selectively using a plurality of communication standards. For example, a user can select which of LTE, wireless LAN, and Bluetooth (registered trademark) to use with a smart phone. It is also possible to selectively use lines of different carriers using the same communication standard.

Since each access network has different communication qualities such as bandwidth and delay, these must be used appropriately according to the application. As shown in FIG. 1, the user can manually switch the access network to be used by changing the settings of the user terminal. However, when there are many available networks, a difficulty arises that the user needs to understand the characteristics of each access network and manually select and set the access network appropriately.

If a user terminal connects to an access network that does not have the desired communication quality or is not suitable for its purpose, the communication quality of other user terminals that use the access network will also deteriorate. Thus, there is a risk that effective utilization of communication resources in the entire network system cannot be achieved.

For example, when the user is connected to an access network that does not have the desired communication quality, the following situation occurs.
If you try to use an access network with insufficient radio wave strength, such as a public wireless LAN at a station, etc., the degree of modulation will be lowered, so it is necessary to allocate more communication resources such as time and frequency to the user. There is In other words, the communication quality of other users is significantly degraded due to the allocation of communication resources.

Also, when connecting to an access network that does not meet the user's purpose, the following situation occurs.
Suppose there exists a network (a) with a small total bandwidth but low delay. If a user who uses an application that does not require low latency, such as video viewing, uses a large amount of the bandwidth of network (a), communication delay increases for users who use applications that require low latency, such as online games. and satisfaction will be greatly reduced. That is, other users may be dissatisfied by being connected to an access network that is not fit for purpose.

For the above difficulties, it is known that a user terminal has a function of automatically selecting an access network (see, for example, Patent Document 1). This function predicts the communication quality from the radio wave intensity of an available wireless LAN line, etc., and preferentially connects to the wireless LAN line when sufficient quality can be expected and is available.

This function estimates communication quality from radio access information. For this reason, an error may occur between the estimated value and the actual communication quality depending on the degree of congestion of the upper network and the behavior of other user terminals. In order to obtain the actual communication quality, there is a problem that it is necessary to connect to the line once and measure the quality.

In addition, since this function cannot control the connection destinations of other user terminals, it is difficult to optimize the network for the entire system. For example, even if there is a user 1 who wants to use an application that requires low-delay communication, the network 2 that allows low-delay communication is already occupied by the terminal of the user 2 who is using an application that does not require a strict delay. Consider the case when you are stuck. In this case, a delay occurs in user 1's communication, making it difficult to achieve user 1's satisfaction. On the other hand, the quality is excessive for user 2's communication. Thus, it is difficult to optimize the service with the function of Patent Document 1.

On the other hand, in order to solve the problem of Patent Document 1 (optimize the entire network system), there is a method in which a server or base station device on the network instructs a user terminal to connect to (for example, non-patent document 1). FIG. 3 is a diagram for explaining the method of Non-Patent Document 1. As shown in FIG. By using this method, it is possible to collectively control the connection destinations of a plurality of users in consideration of the degree of congestion of the entire network. Non-Patent Document 1 can improve the throughput of the entire system in an environment where 3GPP lines and wireless LAN lines coexist.

JP 2012-169971 A

D. Kimura, “A Novel RAT Virtualization System with Network-Initiated RAT Selection Between LTE and WLAN”, 978-1-5090-4183-1/17, 2017 IEEE Ouya Ono, Kiyoshi Narukawa, “Proposal of optimum access automatic selection method adapted to multi-access environment”, 2019 Institute of Electronics, Information and Communication Engineers Society Conference, B-6-36 https://business. ntt-east. co. jp/service/azukeru_ms/, retrieved on December 31, 2019 https://www. ntt. com/business/services/application/mail-groupware/office365. html, retrieved on December 31, 2019

The network-based connection destination selection algorithm of Non-Patent Document 1 is a binary algorithm including switching between LTE and wireless LAN. In other words, Non-Patent Document 1 has the first problem of difficulty in expandability to an environment in which a wider variety of access networks can be used.

The selection algorithm of Non-Patent Document 1 has the second problem that the objective function of optimization is only a variable of throughput, and it is difficult to improve the user satisfaction of applications that emphasize indicators other than throughput. In recent years, some applications have appeared in which delay and delay fluctuation have a large impact on user satisfaction. Therefore, connection destination selection methods that consider only throughput cannot sufficiently improve user satisfaction.

In addition, each communication standard has physical properties determined by radio frequencies and unique feature amounts determined by service forms such as cost. The selection algorithm of Non-Patent Document 1 does not consider these feature amounts, and selects connection destinations without reflecting the features of each access network. In other words, Non-Patent Document 1 cannot select a connection destination considering the characteristics of each access network, and in this respect also has the third problem that user satisfaction cannot be sufficiently improved.

As a means of solving the above three problems, a means of optimizing a user's access destination based on a utility function defined by taking into account multiple communication quality parameters and feature values for each means of access has been disclosed ( See Non-Patent Document 2).

On the other hand, conventional data communication has centered on Internet communication, but in recent years, demand for cloud services and low-delay applications has increased. In order to meet such demands, there are cases where an application server is installed in the network of a communication carrier and provided as a service together with a communication line (for example, see Non-Patent Documents 2 and 3). In such a case, it is required to limit the network to use the application or enjoy the service, or to connect to a specific network to improve the quality of experience of the application.

However, Patent Document 1 and Non-Patent Documents 1 and 2 are methods for optimizing communication paths from the upper network to the Internet, and the above (1) application that can be used only under a specific network (2) under a specific network It does not take into consideration applications that increase the quality of experience in Therefore, Patent Literature 1 and Non-Patent Literatures 1 and 2 have a problem that it is not always possible to connect to an access route desired by a user who intends to use the application.

Therefore, in order to solve the above problems, the present invention provides an optimization engine that can improve the selection accuracy of a network suitable for an application and improve the quality of experience without making major changes to the communication system. It aims at providing a method and a program.

To achieve the above object, an optimization engine according to the present invention comprises an objective function of items to be improved, collects parameters from access networks and terminals, and has a destination server installed in a specific access network. With this in mind, we decided to find a combination of connection destinations that maximizes or minimizes the objective function.

Specifically, the optimization engine according to the present invention is an optimization engine for a communication system, comprising:
The communication system has a configuration in which each of a plurality of terminals is connected to an upper network via one of a plurality of access networks,
The optimization engine is
an information aggregating unit that collects communication quality information and network feature values for each of the access networks, and usability information indicating which of the access networks can be used for each of the terminals;
a candidate selection unit that creates connection destination candidates, which are candidates for the access network to which each of the terminals connects, based on the availability information;
a quality estimating unit that estimates the communication quality of the connection destination candidate based on the communication quality information and sets the communication quality as an estimated communication quality;
a determination unit that determines an optimum connection destination from among the connection destination candidates based on a calculated value obtained by substituting the network feature amount and the estimated communication quality into a preset objective function;
and
If there is a correspondence relationship between an application and a specific access network that can implement the application among the access networks,
When estimating the communication quality, the quality estimating unit determines whether the terminal uses the application via the specific access network or when the terminal uses the application via another access network. is characterized by performing a process of causing a difference in the communication quality.

Further, an optimization method according to the present invention is an optimization method for a communication system,
The communication system has a configuration in which each of a plurality of terminals is connected to an upper network via one of a plurality of access networks,
The optimization method includes:
Collecting communication quality information and network feature values for each of the access networks, and usability information indicating which of the access networks can be used for each of the terminals;
creating connection destination candidates, which are candidates for the access network to which each of the terminals connects, based on the usability information;
estimating the communication quality of the connection destination candidate based on the communication quality information as an estimated communication quality;
Determining an optimum connection destination from among the connection destination candidates based on a calculated value obtained by substituting the network feature quantity and the estimated communication quality into a preset objective function;
is doing
If there is a correspondence relationship between an application and a specific access network that can implement the application among the access networks,
When estimating the communication quality, there is a difference in the communication quality between when the terminal uses the application via the specific access network and when the terminal uses the application via another access network. It is characterized by performing a process that causes

This optimization engine and its method select a combination of connection destinations from among a plurality of access networks based on an objective function having a plurality of communication quality parameters and a plurality of network feature quantities as variables. By appropriately setting the objective function, it is possible to control the bandwidth usage rate, line usage cost, and the like. Using values that can be obtained from network devices and user terminals, or values that can be derived using them, it is possible to derive connection destinations of user terminals according to an arbitrary objective function.

Also, when estimating the communication quality used for the objective function, the optimization engine and its method estimate the communication quality by a network simulator or based on the setting of the network equipment. At that time, the network device is configured so that the calculated value changes depending on whether the server is accessed via the access network to which the specific server is connected or when the server is accessed via another access network. Make a pseudo setting for

Therefore, the present invention can provide an optimization engine and an optimization method that can improve the accuracy of selecting a network suitable for an application and improve the quality of experience without making major changes to the communication system.

For example, the connection destination candidate having the maximum or minimum value of the objective function can be set as the optimum connection destination.

The optimization engine according to the present invention is characterized by further comprising a notification unit that outputs a connection command to each of the terminals and the access network so that the connection between the terminal and the access network becomes the optimum connection destination. do.

Further, the optimization method according to the present invention further comprises outputting a connection command to each of the terminals and the access network so that the connection between the terminal and the access network becomes the optimum connection destination. Characterized by

A program according to the present invention is a program for causing a computer to function as the optimization engine. The optimization engine according to the present invention can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided through a network.

INDUSTRIAL APPLICABILITY The present invention can provide an optimization engine, an optimization method, and a program capable of improving the selection accuracy of a network suitable for an application and improving the quality of experience without making major modifications to a communication system. .

It is a figure explaining the communication system relevant to this invention. It is a figure explaining the communication system relevant to this invention. It is a figure explaining the communication system relevant to this invention. 1 is a diagram illustrating a communication system comprising an optimization engine according to the invention; FIG. 1 is a diagram illustrating a communication system comprising an optimization engine according to the invention; FIG. 1 illustrates a communication system comprising an optimization engine according to the invention; FIG. It is a figure explaining operation|movement of the optimization engine which concerns on this invention. FIG. 3 is a diagram for explaining the operation of a communication system having an optimization engine according to the present invention; 1 is a diagram illustrating a communication system comprising an optimization engine according to the invention; FIG. 1 is a diagram illustrating a communication system comprising an optimization engine according to the invention; FIG. It is a figure explaining the optimization engine based on this invention.

Embodiments of the present invention will be described with reference to the accompanying drawings. The embodiments described below are examples of the present invention, and the present invention is not limited to the following embodiments. In addition, in this specification and the drawings, constituent elements having the same reference numerals are the same as each other.

(Embodiment 1)
[Communications system]
FIG. 4 is a diagram illustrating a communication system 301 that includes the optimization engine 50 of this embodiment. A communication system 301 is configured such that each of a plurality of terminals 11 is connected to an upper network 13 via one of a plurality of access networks (NW) 12 . The communication system 301 in FIG. 4 has four terminals 11 and four NWs 12, but these numbers are not limited to four.

The optimization engine 50 calculates an objective function such as a user satisfaction function using a plurality of communication quality parameters and a plurality of network feature values as variables from among the plurality of NWs 12, and connects the terminal 11 and the NW 12 respectively. Select combinations dynamically. "Dynamic selection" means that the objective function is calculated periodically and the connection combination is switched according to the result.
The communication quality parameters are the total NW bandwidth, delay, delay fluctuation, number of available TCP sessions, number of available IP addresses, and other communication quality parameters.
The network feature amount is a value indicating line usage cost, resistance to movement of the user (mobility), presence/absence of encryption, and other features of the network.
If the communication quality parameters and network feature values are values that can be obtained from network devices and user terminals, or values that can be derived using them, the NW 12 to which the terminal 11 should be connected is derived according to an arbitrary objective function.

The optimization engine 50 can obtain the following effects.
(1) Appropriate NW 12 to be used by terminal 11 can be selected in an environment where many NWs 12 exist.
(2) Even when the number of available NWs 12 increases, the destination selection algorithm can be easily extended.
(3) The terminal 11 can select the NW 12 that provides a higher level of satisfaction even for an application in which a plurality of parameters other than throughput are related to the level of satisfaction.
(4) The terminal 11 and the NW 12 can be connected in consideration of the characteristics of each NW 12 .
(5) By designing the objective function, it is possible to realize connections between the terminals 11 and the NW 12 that meet various demands, such as maximizing user satisfaction and averaging the load factor of each network. In other words, by setting a desired objective function in the optimization engine 50, it is possible to control the overall communication system 301 in consideration of the band utilization rate of the NW 12, line usage cost, and the like.
[supplement]
The "connection destination selection algorithm" is a series of procedures for selecting the NW connection destination selected by the user terminal (described later, after the objective function setting procedure, the search candidate selection, quality estimation, and objective function evaluation procedures (search loop) repeatedly).
"Easy expansion of connection destination selection algorithm" means that there is no need to change the above series of procedures or objective functions, and mainly by expanding the function of the quality estimation unit, the type and number of available NWs can be changed, Or it means that it can cope with fluctuations in the number of terminals. The reason why the connection destination selection algorithm can be easily expanded is that the functions of objective function setting, search candidate selection, quality estimation, and objective function evaluation, which will be described later, are highly independent, and function expansion is easy. be. In other words, when the number of usable NWs increases, it can be dealt with only by changing a part of the functions, and the objective function and the flow chart do not need to be changed significantly.
It should be noted that, as in Non-Patent Document 1, many of the conventional connection destination selection algorithms are alternative methods of 3GPP line and wireless LAN line, and significant algorithm renewal is required to apply to three or more NWs. is necessary. In addition, such a method directly reflects the characteristics and relationship of the 3GPP line and the wireless LAN line in the function of selecting the connection destination. ), the introduction of a new NW requires reconstruction of the connection destination selection function.

FIG. 5 is a block diagram illustrating the functionality of terminal 11, access network 12, and optimization engine 50. As shown in FIG.
The terminal 11 has a terminal information notification unit 11a that notifies the optimization engine 50 of the application to be used and the NW 12 that can be used.
The terminal 11 has a network selector 11b that receives instructions from the optimization engine 50 and switches the NW 12 to be used.

The NW 12 has a terminal selector 12a that receives instructions from the optimization engine 50 and switches the terminal 11 to be connected. Either one of the terminal selection unit 12a of the NW 12 and the network selection unit 11b of the terminal 11 may be used, or both may be used simultaneously.
The NW 12 has a network information notification unit 12b that notifies the optimization engine 50 of its own communication quality information such as available bandwidth.

The optimization engine 50 has an information aggregation unit 51 that aggregates information from the information notification unit 11 a of the terminal 11 and the network information notification unit 12 b of the NW 12 .
The optimization engine 50 has a search candidate selection unit 52 that defines a set of connection combinations between the terminal 11 and the NW 12 and extracts connection combination candidates at the time of searching from the set.
The optimization engine 50 has a quality estimator 53 that simulates or estimates quality by simulating the real world. The quality estimating unit 53 outputs the estimated quality of each terminal 11 when the connection combination candidates are input and connected to them.
The optimization engine 50 has an objective function evaluation unit 54 that calculates the value of the objective function based on the communication quality of each terminal 11 and the like.
The optimization engine 50 has an evaluation result determination unit 55 that receives the calculation result of the objective function evaluation unit 54 and determines whether to search for connection combinations again or end the search.
The optimization engine 50 has an optimum network notification unit 56 that notifies at least one of each terminal 11 and each NW 12 of the finally determined connection combination.

The optimization engine 50 is
The information aggregating unit 51 collects communication quality information (“P” to be described later) and network feature amount (“C” to be described later) for each NW 12, and usability information (described later) as to which NW 12 can be used for each terminal 11. "A") of
The search candidate selection unit 52 creates connection destination candidates (connection combinations), which are candidates for the NW 12 to which each of the terminals 11 is connected, based on the availability information,
a quality estimating unit 53 estimating the communication quality of the connection destination candidate based on the communication quality information (outputting the estimated communication quality);
The objective function evaluation unit 54 substitutes the network feature quantity and the estimated communication quality into a preset objective function,
The evaluation result determination unit 55 determines the connection destination candidate having the maximum value or the minimum value of the objective function as the optimum connection destination.

Then, the optimum network notification unit 56 outputs a connection command to at least one of the terminals 11 and NW 12 so that the connection between the terminals 11 and NW 12 becomes the optimum connection destination.

[motion]
FIG. 6 is a diagram for explaining the operation of the optimization engine 50. As shown in FIG. It is assumed that each terminal 11 can use a maximum of N types of NWs 12 within the communication system. Assume that an array of values of communication quality parameters such as the total bandwidth and average delay of the n-th NW 12 is represented by a vector _Pn . It is also assumed that the array of feature quantities other than communication quality possessed by the n-th NW 12 is represented by a vector _Cn . The number of elements of vector _Pn and vector _Cn is equal to the number of communication quality parameters and feature quantities to be considered. A set of vectors _Pn is set P, and a set of vectors _Cn is set C.

Note that vector C _n may contain a list of applications implemented in the nth NW 12 .

There are M terminals 11 in the communication system. Let the array indicating the NWs 12 available to the m-th terminal 11 be a vector _Am , and let the set of the vectors _Am be a set A. The vector _Am describes the number of available NWs 12, and its number of elements is equal to the number of available NWs 12. Alternatively, the vector A _m may be defined as an array with the number of elements N such that the i-th component A _m,i is as follows.

The vector Am may include application information (application information) used by the m-th terminal 11 .

FIG. 7 is a diagram for explaining the operation of the optimization engine 50. As shown in FIG.
A combination of NWs 12 to which M terminals 11 are connected is defined as a vector x, and the mth element _xm indicates the number of the NW 12 to which the mth terminal 11 is connected ( _1≤xm≤N ).
The search candidate selection unit 52 of the optimization engine 50 receives as input the available access network A, which is a set of vectors _Am representing NWs 12 that can be used by each terminal 11, and generates a set X of connection destination candidates for each terminal.

The vector x is a vector in which the connection destination NW numbers _xm to which the m-th terminal is connected are arranged for all terminals. A vector x may be described as a "connection combination". On the other hand, vector A _m is a vector in which NW numbers that can be used by the m-th terminal (connection candidates) are arranged.

Then, the search candidate selection unit 52 generates a set X including connection destination candidates of all terminals using the set A of vectors _Am of all terminals. Furthermore, the search candidate selection unit 52 selects one connection candidate for each terminal from the set X as the first (i=1) search loop, arranges them, and sets the quality as the connection destination candidate x ₁ of the vector. Input to the estimation unit 53 . The search candidate selection unit 52 changes at least one connection candidate of the terminal each time the search loop is performed, and sets it as a connection destination candidate _xi of a new vector. That is, the connection combination x changes for each search loop.

The quality estimation unit 53 receives the connection destination candidates x _i from the search candidate selection unit 52 and calculates the communication quality y _i of the entire terminal. The communication quality y _i is a vector and a function of the connection destination candidate x _i (y _i (x _i )). The m-th component _{yi ,m} of the communication quality yi is the communication quality obtained at the m-th terminal in the connection combination in search loop i.

Assume that the objective function set in the objective function evaluation unit 54 is f(y _i (x _i ), C). Let f ^* be the maximum or minimum value of the objective function obtained up to the i-th search loop. Let x ^* be a connection destination candidate when f ^* is obtained. The objective function evaluation unit 54 calculates the following equation using the feature amount set C obtained from the information aggregating unit 51, the communication quality y _i (x _i ), and the objective function, and outputs the connection destination candidate x ^* .

FIG. 8 is a flow chart for explaining the operation of the communication system 301. As shown in FIG.
The network information notification unit 12b of each NW 12 notifies the information aggregation unit 51 of the optimization engine 50 of the communication quality information (vector P _n ) and the feature amount (vector C _n ) other than the communication quality. Note that if there are applications implemented within the NW 12, a list of those applications may also be notified. The information aggregation unit 51 creates a set P and a set C from the notified communication quality information and feature amount.
The information notification unit 11a of each terminal 11 notifies the information aggregation unit 51 of the optimization engine 50 of the access network that can be used at any time and the application that is being used or is scheduled to be used (vector A _m ) ( step S02). The information aggregator 51 creates available applications (set A) from the notified applications.

The information notifications in steps S01 and S02 may be performed in any order, or may be performed simultaneously. There are no restrictions on the order of terminals or the order of networks. Also, if the state of the terminal and network is known and dynamic changes do not occur, the settings may be set in advance without performing these steps.

The search candidate selection unit 52 of the optimization engine 50 uses the set A to generate a set X of terminal connection destination combinations (step S03).
The search candidate selection unit 52 initializes the number i of search loops for a solution, the maximum or minimum value f ^* of the value of the objective function, and the connection destination x ^* at that time (the initial value is 0 or a zero vector).
Steps S03 and S04 may be performed in any order and may be performed simultaneously.

The search candidate selection unit 52 of the optimization engine 50 extracts the element _xi from the set X of connection destination candidates. The quality estimator 53 uses the set P to estimate the communication quality _{y i} realized when the element x _i is input (steps SS05 to S08). A method of extracting the elements _xi includes a method of randomly extracting from X or a method of extracting all the elements in a specific order. Furthermore, a method of randomly creating elements x _i and performing quality estimation only for elements that have been confirmed as x _i εX may be used (step S06).

Any method may be used for estimating the communication quality _yi (step S07). For example, the method includes a method of outputting the results of simulation using a system that simulates the actual network and user distribution. By using network simulators such as “ns-3”, “QualNet”, “OpNet Modeler”, etc., it is possible to estimate communication delay and the like in addition to throughput. Also, if only the throughput is to be estimated, there is also a simple estimation method such as dividing the total bandwidth of each network by the number of people connected to that network.
As an example of the estimation method, in the second embodiment, means for estimating the communication quality _yi based on the NW simulator, the NW topology, or the setting of the NW equipment will be described.

The objective function evaluation unit 54 uses a previously given objective function f(y _i , C), and from the communication quality y _i obtained for the connection destination candidate element x _i and the feature amount C other than the communication quality, , the value of the objective function is calculated (step S10). For example, when the objective function f(y _i , C) includes QoE (Quality of Experience) representing user satisfaction, the QoE model (step S09) determined from the application used by each terminal obtained in step S02 is used. , the communication quality y _i and the feature amount C, the QoE value is derived. Specifically, assuming that web browsing is performed as an application, the QoE can be estimated from the communication quality _yi by using the QoE model for web browsing, the required bandwidth of the web page, and the average throughput. .

The evaluation result determination unit 55 determines that the objective function f(y _i , C) is larger than the maximum value f ^* obtained in the search so far (calculation up to the number of search loops i−1) at the number i of search loops. If it is larger ("Yes" in step S10), f ^* is updated to the value of the objective function f(y _i , C) at the search loop count i. Also, the element x _i when the f ^* is obtained is updated as x ^* (step S11).
If the objective function is a function to be minimized, f ^* is updated if f(y _i , C) is smaller than the minimum value f ^* obtained in previous searches (“Yes” in step S10). . Also, the element x _i when the f ^* is obtained is updated as x ^* (step S11).

The search loop for the solution ends n times, which is the number of elements of X, or when a predetermined search end condition is met ("Yes" in step S12). On the other hand, if the search end condition is not met, the search loop is repeated from step S05 ("No" in step S12).
Specifically, search end conditions include the following.
(1) Number of searches (upper limit of i)
(2) Search time (3) The value of a specific index or objective function exceeds or falls below a certain value (4) It is clear that x ^* obtained during the search will not be updated in the future.

When the search is finished, the optimum access network notification unit 56 of the optimization engine 50 notifies at least one of the network selection unit 11b of each terminal 11 and the terminal selection unit 12a of each NW 12 of the connection destination based on x ^* . (steps S13, S14). The terminal 11 and NW 12 that have received the notification switch connection destinations according to the notification. After the respective terminals 11 and NW 12 switch their connection destinations, they enter the state of x ^* .

(Embodiment 2)
In the present embodiment, a method of estimating the communication quality _yi based on the NW simulator or the NW topology and the setting of the NW equipment, which was explained in step S07 of FIG. 8, will be explained.
9 and 10 are diagrams illustrating a communication system 302 that includes the optimization engine 50 of this embodiment. A communication system 302 has an application server (application D) on a specific network (NW12-4). The communication system 302 differs from the communication system 301 in FIG. 4 in that an application server also exists in the NW12. In this embodiment, differences between the communication system 302 and the communication system 301 will be described.

The optimization engine 50 of the communication system 302 selects the access means to be used by the user, taking into consideration that the communication destination application server is installed in a specific NW. In other words, in the case of the communication system 302, by connecting the terminal 11-3 using the application D to the NW 12-4 as shown in FIG. 9, it is possible to suppress the deterioration of the user's QoE and improve the QoE.

However, as shown in FIG. 10, the server of application D is currently not connected to NW12-4. Application D currently exists on the Internet, but application D may be connected to NW 12-4 in the future in order to suppress deterioration in the user's quality of experience and improve the quality of experience. In such a case, normally, it is necessary to use a simulator or the like to simulate the server 25 of the application D connected to the NW 12-4 to estimate the communication quality _yi . However, such a technique requires various alterations other than the quality estimator 53 of the optimization engine 50 .

Therefore, in the present embodiment, a method for simplifying simulation settings while performing highly accurate quality prediction without requiring modifications other than the quality estimation unit 53 of the optimization engine 50 will be described.
If there is a specific correspondence between application D and the specific access network (NW12-4),
When estimating the communication quality _yi , the quality estimation unit 53 determines whether the terminal 11 uses the application D via the specific access network and when the terminal 11 uses the application D via the other access network. and perform processing to cause a difference in communication quality _yi .

That is, when performing step S07 in FIG. 8, the quality estimation unit 53 estimates the communication quality y _i by changing the setting of the NW simulator or the NW topology or the NW equipment. Specific setting changes will be described later. By changing the setting, a situation is created as if the server of application D is connected to NW 12-4 as shown in FIG. 10, and the communication quality y _i is estimated under this situation. By using the communication quality y _i estimated in this way as the objective function (step S10), the probability of connecting the terminal 11-3 that intends to use the application D to the NW 12-4 can be increased.

Note that the existence of the application D may be set in the optimization engine 50 in advance by the operator, or may be received by the information aggregating unit 51 as NW information (feature amount C).

The advantages of this embodiment are as follows.
It is possible to improve the accuracy of selecting a NW suitable for an application service, suppress the deterioration of the quality of experience, and improve the quality of experience. In addition, the communication quality estimation method of Embodiment 1 can be utilized as it is (just by adding a utility function to the objective function) without modification.

ただし、ｈ_ｍはユーザｍの満足度推定値である。
（２）満足度推定値が設定値以上になるユーザの人数（最大化）

ただし、ｈ_ｏは定数である。
（３）満足度推定値の中央値（最大化）

（４）満足度推定値の分散（最小化）

（５）満足度推定値が最低のユーザの満足度（最大化）

（６）ネットワークの負荷率の偏り（最小化）

ただし、ｌ_ａとｌ_ｂは、それぞれアクセスネットワークＡとＢの帯域利用率である。
なお、アクセスネットワークが３つ以上存在する場合はそれぞれのアクセスネットワークの帯域利用率ｌ_ｎの分散を用いればよい（本式でｎはアクセスネットワークの識別番号）。

（７）所望のネットワーク負荷率配分からの差異（最小化）

ただし、γは、ｌ_ａ：ｌ_ｂをａ：ｂとしたい場合、ｂ／ａである。“ａ”及び“ｂ”は、正の数である。例えば、ネットワークＡとネットワークＢの帯域利用率を、ｌ_ａ：ｌ_ｂ＝３：４の比になるよう接続先を振り分ける場合、ａ＝３、ｂ＝４とする。
なお、アクセスネットワークが複数（Ｎ個）存在する場合、数Ｐ７を一般化すると次のような目的関数となる。

ただし、ｌ_０とｌ_ｎは、それぞれ基準となるアクセスネットワークとｎ番目のアクセスネットワーク（ｎはＮ以下の整数であって、基準のアクセスネットワークを除く）の帯域利用率である。γ_ｎは、基準となるネットワークとn番目のネットワークの帯域利用率の比の値である。つまり、γ_ｎは、基準となるネットワークの帯域利用率ｌ_０とｎ番目のネットワークの帯域利用率ｌ_ｎをｌ_０：ｌ_ｎ＝１：γ_ｎとする場合の値である。
（８）仮想移動体通信事業者（ＭＶＮＯ：Ｍｏｂｉｌｅ Ｖｉｒｔｕａｌ Ｎｅｔｗｏｒｋ Ｏｐｅｒａｔｏｒ）から見た回線利用コストの合計値（最小化）

ただし、ｐ_ｎはアクセスネットワークｎの帯域あたりの回線利用料、Ｂ_ｎはアクセスネットワークｎの単位時間当たりの利用データ量である。[Specific example of objective function]
The objective function setter determines the objective function in consideration of the business model, user satisfaction, fairness, cost, or the like. By setting the objective function f(y, C), it is possible to sort the connection destinations of users according to various policies.
Examples of objective functions are given below.
(1) Sum of estimated satisfaction values (maximization)

where h _m is the satisfaction estimate for user m.
(2) Number of users whose estimated satisfaction value is greater than or equal to the set value (maximization)

However, _ho is a constant.
(3) median satisfaction estimate (maximization)

(4) Variance (minimization) of satisfaction estimate

(5) User Satisfaction with Lowest Satisfaction Estimate (Maximize)

(6) Imbalance in network load factor (minimization)

where l _a and l _b are the bandwidth utilization rates of access networks A and B, respectively.
If there are three or more access networks, the variance of the band utilization rate _ln of each access network may be used (where n is the identification number of the access network).

(7) Deviation from desired network load factor distribution (minimization)

However, γ is b/a when l _a : l _b is desired to be a:b. "a" and "b" are positive numbers. For example, when distributing the connection destinations so that the ratio of bandwidth usage between network A and network B is l _a : l _b =3:4, a=3 and b=4.
When there are a plurality (N) of access networks, generalizing the number P7 results in the following objective function.

where l ₀ and l _n are the bandwidth utilization rates of the reference access network and the n-th access network (n is an integer less than or equal to N, excluding the reference access network), respectively. _γn is the value of the ratio of the bandwidth utilization rates of the reference network and the n-th network. That is, γ _n is a value when the band utilization rate l ₀ of the reference network and the band utilization rate l _n of the n-th network are l ₀ : l _n =1: γ _n .
(8) Total value (minimization) of line usage cost seen from a virtual mobile network operator (MVNO: Mobile Virtual Network Operator)

However, _pn is the line usage fee per bandwidth of the access network n, and _Bn is the amount of data used per unit time of the access network n.

When setting objective functions of numbers P1, P2, P3, and P5, optimization is performed to maximize them in step S10 of FIG. On the other hand, when setting objective functions of numbers P4, P6, P7, and P8, optimization is performed to minimize them in step S10 of FIG.

と定義する。In addition, the user satisfaction used in the numbers P1 and P2 is the QoE (Quality of experience) determined by the communication quality of each application, and the influence α( It is an index that considers both of C). in particular,

defined as

When it is desired to improve a plurality of objective functions f ₁ (y, C), f ₂ (y, C), . If the degree of emphasis on the objective function f _j (y, C) differs, each f _j (y, C) can be weighted (j is a natural number). An example of a method of synthesizing objective functions will be described below. where the vector w is an array of weighting percentages.

(A) Expressing as a sum

(B) Expressing by product

(C) Method of Replacing with Another Function and Representing it as a Sum or Product You may not be able to think in a balanced way. For example, if an objective function to be maximized and an objective function to be minimized are to be considered at the same time, it is necessary to perform transformation such as taking the reciprocal of one of the functions before combining them. In addition, if there is a difference in the range of values that can be taken by each objective function, there is a possibility that the contribution of the objective function with a large range of values will increase. Therefore, if it is desired to equally consider a plurality of objective functions, it may be necessary to perform normalization to adjust the possible range. It is also possible to express the objective function f by using g as a function used for these transformations and using a composite function of g and _fj .

For example, the objective function f can be set as follows.

とすればよい。最大化すべき関数から最小化すべき関数へ変換を行いたい場合にも同様の関数で変換を行えば良い。Here, if you want to convert from a function that should be minimized to a function that should be maximized,

And it is sufficient. A similar function may be used to convert from a function to be maximized to a function to be minimized.

ただし、ａ及びｆ_０は定数である。Also, the sigmoid function of Equation 10 can be set as an example of the function g that converts the range of possible values. Then g(f _j (y, C))ε[0, 1] for any value of f _j (y, C).

where a and _f0 are constants.

ただし、ａ及びｆ_０は定数である。Also, Equation 11 can be set as another example of the function g. In this case, g(f _j (y, C))ε[0, 1] under the condition that f _j (y, C)≧0.

where a and _f0 are constants.

とすればよい。また、３つ以上の関数の合成関数としてもよい。Multiple functions g _k and g _l may be used simultaneously if multiple transformations are required. The function g finally used at that time is

And it is sufficient. Alternatively, a composite function of three or more functions may be used.

[Specific example of setting change]
A specific example of a technique for artificially reproducing a situation in which use of an application from an external NW is restricted will be described.
1. When an application can be used in a specific NW (1) Method of simulated reproduction by simple setting (i) Use of firewall function It may be assumed that the server of application D is on the Internet on the simulator. The server receives only packets from the IP address (terminal 11) under NW 12-4 by the firewall function provided to the server. Alternatively, a similar firewall function is installed in NW equipment present on the communication path from the terminal 11 to the server.
That is, only terminals under NW 12-4 can communicate with the server, and terminals under other NW 12 (external NW) cannot communicate with the server, resulting in a difference in estimated communication quality _yi .
(ii) Routing setting Static routing is set to the router existing on the communication path from the terminal 11 to the application D, and packets addressed to the server of the application D received from a specific port or IP address (connection to the external NW) packet from the terminal) is not forwarded or is forwarded to another route. As an example of the installation location of the router to be set, the boundary between the upper network 13 and the Internet 20 and the front of the application D can be considered. Alternatively, in the server, the setting of the default gateway addressed to a specific IP address is set differently from the original route.
In other words, the routing of packets from a terminal connected to an external NW makes it impossible to transfer packets or reduces the throughput, resulting in a difference in the estimated communication quality _yi .
(iii) Communication interruption, communication on a different route A link with an infinitely small bandwidth or a sufficiently large delay is generated in the route from the upper network 13 to the server of the application D, and IP address communication from the external NW is prohibited. Form a topology that passes through that link. Alternatively, the QoS value is set low for packets destined for the external NW so that transfer in switches and routers is not performed with sufficient quality.
In other words, the throughput and quality of packet transfer from a terminal connected to the external NW are reduced, so that the estimated communication quality _yi varies.

2. In the case of an application whose utility increases in a specific NW: (1) Method of pseudo-reproduction by simple setting (i) Blocking of access from external NW The same method as the use of the firewall function described above is performed.
(ii) Routing setting The same method as the routing setting described above is performed.
(iii) Decreasing the quality of access from external NW It may be assumed that the server of application D is on the Internet on the simulator. A NW topology is set in which access to the server from the external NW passes through a communication path with a large delay or a communication path with a small bandwidth, or the setting of the link is changed.
In other words, the throughput of packet transfer from a terminal connected to an external NW is reduced by inserting a delay or adjusting a band, so that a difference occurs in the estimated communication quality y _i .

(Example of optimization engine)
The optimization engine 50 described above can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided through a network.
FIG. 11 shows a block diagram of system 100 . System 100 includes computer 105 connected to network 135 . The system 100 corresponds to the communication system 301 and the computer 105 to the optimization engine 50 .

Network 135 is a data communications network. Network 135 may be a private network or a public network, and may be (a) a personal area network covering, for example, a room; (b) a local area network covering, for example, a building; (d) a metropolitan area network covering, e.g., a city; (e) a wide area network covering, e.g. Any or all of an area network, or (f) the Internet. Communication is by electronic and optical signals through network 135 . Note that the network 135 corresponds to the NW 12 and the upper network 13 .

Computer 105 includes a processor 110 and memory 115 coupled to processor 110 . Although computer 105 is represented herein as a stand-alone device, it is not so limited, but rather may be connected to other devices not shown in a distributed processing system.

Processor 110 is an electronic device made up of logic circuitry that responds to and executes instructions.

Memory 115 is a tangible computer-readable storage medium in which a computer program is encoded. In this regard, memory 115 stores data and instructions, or program code, readable and executable by processor 110 to control its operation. Memory 115 may be implemented in random access memory (RAM), hard drive, read only memory (ROM), or a combination thereof. One of the components of memory 115 is program module 120 .

Program modules 120 contain instructions for controlling processor 110 to perform the processes described herein. Although operations are described herein as being performed by computer 105 or a method or process or its subprocesses, those operations are actually performed by processor 110 .

The term "module" is used herein to refer to a functional operation that can be embodied either as a stand-alone component or as an integrated composition of multiple subcomponents. Accordingly, program module 120 may be implemented as a single module or as multiple modules working in cooperation with each other. Further, although program modules 120 are described herein as being installed in memory 115 and thus being implemented in software, program modules 120 may be implemented in hardware (eg, electronic circuitry), firmware, software, or a combination thereof. Either of them can be realized.

Program modules 120 , although shown already loaded into memory 115 , may be configured to be located on storage device 140 for later loading into memory 115 . Storage device 140 is a tangible computer-readable storage medium that stores program modules 120 . Examples of storage devices 140 include compact discs, magnetic tapes, read-only memory, optical storage media, hard drives or memory units consisting of multiple parallel hard drives, and universal serial bus (USB) flash drives. be done. Alternatively, storage device 140 may be random access memory or other type of electronic storage device located in a remote storage system, not shown, and connected to computer 105 via network 135 .

System 100 further includes data source 150 A and data source 150 B, collectively referred to herein as data source 150 and communicatively coupled to network 135 . In practice, data sources 150 may include any number of data sources, one or more. Data sources 150 contain unstructured data and can include social media.

System 100 further includes user device 130 operated by user 101 and connected to computer 105 via network 135 . User device 130 includes input devices such as a keyboard or voice recognition subsystem for allowing user 101 to communicate information and command selections to processor 110 . User device 130 further includes an output device such as a display or printer or speech synthesizer. A cursor control, such as a mouse, trackball, or touch-sensitive screen, allows user 101 to manipulate a cursor on the display to convey further information and command selections to processor 110 . Note that the user device 130 corresponds to the terminal 11 .

Processor 110 outputs results 122 of executing program modules 120 to user device 130 . Alternatively, processor 110 may provide output to storage 125, such as a database or memory, or via network 135 to a remote device not shown.

For example, the program module 120 may be a program that performs the operation of FIG. System 100 may operate as optimization engine 50 .

The terms “comprising” or “comprising” specify that the feature, entity, step, or component recited therein is present, but one or more It should not be construed as excluding the presence of other features, integers, steps or components, or groups thereof. The terms "a" and "an" are indefinite articles and thus do not exclude embodiments having a plurality thereof.

(Other embodiments)
It should be noted that the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the gist of the present invention. In short, the present invention is not limited to the high-level embodiments as they are, and can be embodied by modifying the constituent elements without departing from the scope of the present invention at the implementation stage.

Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. For example, some components may be omitted from all components shown in the embodiments. Furthermore, constituent elements across different embodiments may be combined as appropriate.

11: Terminal 11a: Terminal information notification unit 11b: Network selection unit 12: Access network (NW)
12a: Terminal selection unit 12b: Network information notification unit 13: Upper network 20: Internet 25: Simulated server 50: Optimization engine 51: Information aggregation unit 52: Search candidate selection unit 53: Quality estimation unit 54: Objective function evaluation Unit 55: Evaluation result determination unit 56: Optimal network notification unit 100: System 101: User 105: Computer 110: Processor 115: Memory 120: Program module 122: Result 125: Storage device 130: User device 135: Network 140: Storage device 150: data source 301, 302: communication system

Claims (7)

An optimization engine for a communication system,
The communication system has a configuration in which each of a plurality of terminals is connected to an upper network via one of a plurality of access networks,
The optimization engine is
an information aggregating unit that collects communication quality information and network feature values for each of the access networks, and usability information indicating which of the access networks can be used for each of the terminals;
a candidate selection unit that creates connection destination candidates, which are candidates for the access network to which each of the terminals connects, based on the availability information;
a quality estimating unit that estimates the communication quality of the connection destination candidate based on the communication quality information and sets the communication quality as an estimated communication quality;
Based on a calculated value obtained by substituting the network feature quantity and the estimated communication quality into a preset objective function, connection combinations between the plurality of terminals and the plurality of access networks are dynamically determined according to the objective function. a determination unit that selects to
and
If there is a correspondence relationship between an application and a specific access network that can implement the application among the access networks,
When estimating the communication quality, the quality estimating unit determines whether the terminal uses the application via the specific access network or when the terminal uses the application via another access network. , an optimization engine characterized by performing processing that causes a difference in the communication quality. 2. The optimization engine according to claim 1, wherein the determination unit determines the connection destination candidate having the maximum value or the minimum value of the objective function as the optimum connection destination. 3. The apparatus according to claim 2 , further comprising a notification unit that outputs a connection command to at least one of the terminal and the access network so that connection between the terminal and the access network becomes the optimum connection destination. optimization engine. A method of optimizing a communication system, comprising:
The communication system has a configuration in which each of a plurality of terminals is connected to an upper network via one of a plurality of access networks,
The optimization method includes:
Collecting communication quality information and network feature values for each of the access networks, and usability information indicating which of the access networks can be used for each of the terminals;
creating connection destination candidates, which are candidates for the access network to which each of the terminals connects, based on the usability information;
estimating the communication quality of the connection destination candidate based on the communication quality information as an estimated communication quality;
Based on a calculated value obtained by substituting the network feature quantity and the estimated communication quality into a preset objective function, connection combinations between the plurality of terminals and the plurality of access networks are dynamically determined according to the objective function. and selecting to
is doing
If there is a correspondence relationship between an application and a specific access network that can implement the application among the access networks,
When estimating the communication quality, there is a difference in the communication quality between when the terminal uses the application via the specific access network and when the terminal uses the application via another access network. An optimization method characterized by performing processing that causes 5. The optimization method according to claim 4, wherein the connection destination candidate having the maximum value or the minimum value of the objective function is determined as the optimum connection destination. 6. The method according to claim 5 , further comprising: outputting a connection command to at least one of said terminal and said access network so that connection between said terminal and said access network becomes said optimum connection destination. optimization method. A program for causing a computer to function as the optimization engine according to any one of claims 1 to 3.

Images